saved
A Proposal for the Future of Scientific Communication
Hraness cites a source capture. The source author remains the source.
gist
Paradigma proposes Flywheel as a git-like ledger for research projects so AI-era scientific work can be shared as transparent graphs instead of finished PDFs. As a demo, a Codex session with their internal harness and a math checkpoint proved the 2×2 case of Gröchenig’s RH criterion via strict log-concavity of a related curve, fully Lean-formalized without sorry, while stressing it is not a proof of RH. They argue process-visible graphs lower review cost and absorb partial results better than paper-centric peer review.
ideas
- Lottery-ticket science is immature. Buying private compute to dent big problems without shared process is a sign the new paradigm still lacks constructive collection and reuse.
- Flywheel is a research ledger. Like git for software, it makes each step explicit so onboarding, negative results, and frontier contributions stay inspectable.
- A graph is not a paper. Papers claim finished results for scarce reader attention; graphs keep dead ends attached and invite open rebuttal or correction.
- The demo is bounded. Their harness proved the 2×2 Gröchenig tests via strict log-concavity and Lean formalization, not the full Riemann hypothesis.
- Review can be compute-backed. Other models spending compute to break results, with the graph recording what held, is the next peer-review shape they want to invent.
quotes
“You can think of Flywheel like a ledger for a given research project, making each step of the process explicit”
“Publishing a Flywheel graph is not publishing a paper.”
“A graph is a record of work in progress, with the dead ends still attached, offered for review.”
“This is not a proof of RH. Showing the 2×2 case says nothing about 3×3 or larger.”